Considerable gains in life expectancy in the US population have occurred over the past several decades, yet geographic differences in life expectancy have persisted or even increased. The underlying causes of these regional gradients in disease and mortality rates are multiple and complex. A sizable body of evidence has established the role of diet in both prevention and management of chronic diseases; therefore, it is reasonable to posit that diet may be a mediating factor in regional disparities in health. Yet, there is little information in the published literature on geographic variation in dieary attributes or the possible contribution of regional differentials in diet to regional disparities i health outcomes in the US population. An understanding of regional differences in dietary attributes in relation to regional differentials in mortality informs us about pathways through which disparate regional health outcomes are manifested and may be ameliorated. To fill these gaps, we propose to use nationally representative dietary and mortality data collected over the past 20 years (1988- 2010) to examine two related hypotheses. The first hypothesis tests the independent association of region of the country with dietary attributes, patterns, and biomarkers. The specific aims of this hypothesis are to examine geocoded data from the third National Health and Nutrition Examination survey (NHANES III, 1988- 1994) and the NHANES 1999-2010 to determine whether dietary patterns and nutritional biomarkers of American adults (n> 40,000) differ by region of the country, and the nature of 20-year secular trends in this association. The second hypothesis examines whether regional differences in diet patterns and dietary biomarkers are related to regional differences in the prospective risk of mortality after a average of 10 years of follow-up in Americans aged >40 years at baseline. The specific aims of the second hypothesis are to examine nationally representative geocoded individual-level diet and mortality data from the NHANES III (1988-1994) and the NHANES 1999-2004 (n>17,000; >4600 presumed deceased) to determine the extent of regional differences in all-cause and cause-specific mortality that are attributable to regional differences in dietary patterns after adjustment for individual and neighborhood characteristics. We will use fixed effects multiple linear regression and Cox proportional hazards regression models, and multilevel regression modeling with hierarchical random effects suitable for complex survey data to examine associations of dietary factors and their interactions with geographical characteristics.